Informatica used Microsoft Ignite 2025 to lay out a
deeper integration with Microsoft Foundry. The message was simple: AI agents only work well when the data behind them is accurate, trusted, and governed. As companies try to move agentic AI from experiments into real production use, they’re realizing they need cleaner data and better controls. The expanded partnership brings Foundry’s agentic AI tools together with Informatica’s data management system to make that transition easier and more reliable.
Connecting AI Agents to Trusted, Governed Data
A major highlight is the new Informatica MCP Server for Foundry Agent Service. The update creates a direct link between AI agents built in Foundry and Informatica’s Intelligent Data Management Cloud. This allows agents to pull from governed datasets, enforce data quality rules, and understand enterprise-specific semantics in real time.
Rik Tamm-Daniels, GVP Ecosystem Alliances and Technology at Informatica, described the impact of this connection in clear terms to ChannelE2E.
“The core innovation is the Informatica MCP Server for Azure AI Foundry Agents, which allows agents built in Azure AI Foundry to directly connect to Informatica’s metadata system of intelligence and the Intelligent Data Management Cloud using the industry-standard Model Context Protocol,” he said.
This isn’t just a connector - it’s a way to ensure that the data feeding these agents carries the right context and lineage. Tamm-Daniels noted that this grounding allows AI agents to understand industry and business-specific meaning, all while drawing from a governed data pipeline. In practice, that gives partners and customers more confidence that their agents will behave predictably when applied to real operational scenarios.
Streamlining Agentic Development for Partners and MSPs
Informatica also introduced a set of agentic blueprints and a growing library of GenAI recipes for Foundry. These assets are aimed at partners, MSPs, and solution providers who want to operationalize AI without reinventing foundational workflows. The recipes include common patterns, such as multifunction calling and task routing, and industry use cases like loan processing or automotive claims.
According to Tamm-Daniels, these pre-built components are meant to remove early friction in the design and deployment cycle. “The collaboration provides guidance for MSPs and solution providers through pre-built assets and platform interoperability. These recipes are designed to accelerate the configuration, build, connection, management, orchestration, and deployment of Enterprise GenAI solutions,” he said. His point underscores a broader theme at Ignite this year: enterprises aren’t looking for building blocks as much as they’re looking for workable starting points.
This is reinforced by updates to the CLAIRE AI engine, which now integrates with Foundry and Azure OpenAI. CLAIRE brings reasoning and automation across data integration, quality, governance, and MDM. Tamm-Daniels added that “CLAIRE now integrates Azure OpenAI to deliver advanced reasoning across services like data integration, quality, governance, and MDM, helping customers build AI agents with confidence, compliance, and speed.” CLAIRE’s availability across U.S. and European Azure regions also gives customers a clearer compliance path.
These additions dovetail with Informatica’s new support for reading and writing Apache Iceberg tables in Microsoft OneLake. For partners already supporting customers on Microsoft Fabric, this means less time juggling data formats and more time focusing on outcomes. As Tamm-Daniels noted, these integrations “help Microsoft and Informatica partners build and deliver AI agent solutions more easily by streamlining the development cycle and embedding trust by design.”
The message throughout the announcement is consistent: as AI agents move closer to production, the foundation matters. Through these updates, Informatica and Microsoft are giving their joint ecosystem a more controlled, governed, and repeatable way to bring AI into core business operations.
